793975 - with-statement function declaration issue

Status: RESOLVED WORKSFORME

(Core :: JavaScript Engine, defect)

Description

[John-David Dalton]

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_7_4) AppleWebKit/537.1 (KHTML, like Gecko) Chrome/21.0.1180.89 Safari/537.1

Steps to reproduce:

(function() {
  with({}) {
    foo();
    function foo() { alert('hi') }
  }
}());


Actual results:

it errored


Expected results:

alerted "hi"

Comment 1

[John-David Dalton]

I didn't check spec, just looked at existing browser behavior in Chrome 21, Opera 12.02 , and IE 10.

Comment 2

[John-David Dalton]

Related to this Underscore bug: https://github.com/documentcloud/underscore/issues/795

Comment 3

[u279076]

Can you please provide more details? Where are you running that code? What is the error message? What spec does this apply to?

Comment 4

[Boris Zbarsky [:bzbarsky]]

Per spec, the code above is a SyntaxError and should not parse or run at all.  Function declarations are not allowed inside a block; they're only allowed at toplevel in a function or in a toplevel script.

In practice, UAs do various weird stuff to sorta extend the spec here.  Firefox makes it look more or less like this:

  with({}) {
    foo();
    var foo = function foo() { alert('hi') }
  }

which of course throws in your case.  Other UAs hoist the function decl to various places (top of block, top of function, etc); actual interop is poor as with all spec extensions.  So use of constructs like this is ... not suggested.

There are existing bugs about implementing a different behavior here once ECMA actually decides on what that behavior should be and spec it.

Comment 5

[John-David Dalton]

Yap. I thought it might have to do with non-standard function statement jazz (http://es5.github.com/#x12 under "note"). Thanks for confirming.

Comment 6

[u279076]

So is this WONTFIX, INVALID?

Comment 7

[John-David Dalton]

INVALID :D

Comment 8

[Boris Zbarsky [:bzbarsky]]

It's DUPEME of the existing bug on implementing whatever ES6 comes up with here, imo.

Comment 9

[John-David Dalton]

no worries.

Comment 10

[André Bargull [:anba]]

No longer reproducible, probably fixed when the new block-Level Function declaration semantics were implemented. Resolving as WFM.